Diagnostics and Modeling of Plasma Assisted Combustion Kinetics
Igor Adamovich and Walter Lempert
Department of Mechanical and Aerospace Engineering Ohio State University
Seminar at Ecole Centrale, Paris, France, July 3, 2014
Igor Adamovich, Walter Lempert, Sergey Leonov, J. William Rich, and Jeffrey Sutton
Sherrie Bowman, David Burnette, Ben Goldberg, Zak Eckert, Kraig Frederickson, Suzanne Lanier, Ting Li, Munetake Nishihara, Vitaly Petrischev, Annie Pulcini, Andrew Roettgen, Ivan Shkurenkov, Caroline Winters, Zhiyao Yin
OSU NETL group
Plasma assisted combustion: why is it needed?
• Ignition and combustion become unstable at the following conditions: • Low fuel-air (F/A) ratios, or equivalence ratios ϕ= (F/A) / (F/A)stoichiometric : fuel economy and NOx reduction in gas turbines, internal combustion engines, jet engines, unmanned aerial vehicles (UAV)
• Low pressures in combustor (i.e. jet engines at high flight altitudes): preventing engine flameout
• High flow velocities in combustor (i.e. jet engines at high flight speed, scramjet engines)
• Major new capability provided by nonequilibrium plasmas: • Efficient generation of a pool of highly reactive radical species
• Radicals react rapidly with fuel, even at low temperatures
• A wide range of plasma chemical chain branching / fuel oxidation reactions
• This may “nudge” conventional combustion chemistry in the right direction, over a wide range of equivalence ratios, pressures, and flow velocities
Nonequilibrium plasma for ignition, combustion, and flameholding: how does this work?
• High peak reduced electric field (E/N), high electron energy (ε ~ E/N), especially in transient plasmas (nsec pulse duration). Use of short pulse duration also improves plasma stability.
• Significant input energy fraction (tens of %) into inelastic electron impact processes, dissociation and electronic excitation: efficient generation of metastable and radical species (N2*, O*, Ar*, O, H, OH, CH); more radicals generated during N2*, O*, Ar* reactions
• Large pool of radicals enables fuel oxidation at low temperatures, accelerates oxidation at high temperatures
• Effect of plasma-generated radicals on fuel-air flows (over last ~10-20 years): • Reduction of ignition delay at T0 > Tthermal (up to 2-3 orders of magnitude, shock tubes) • Reduction of ignition threshold at T0 < Tthermal (up to 100-200 K, plasma flow reactors) • Increase of flame blow-off velocity (up to a factor of 2, premixed turbulent flames) • Reduction of lean flammability limit (up to ∆ϕ/ϕ ~ 10%, premixed turbulent flames) • Critical for tgnition and flameholding in high-speed flows
• Critical issues / concerns: • Plasma stability at high pressures (discharge filamentation = rapid thermalization) • Scarcity of experimental data at controlled, well-characterized plasma conditions • Validation of kinetic models, developing quantitative predictive capability
PAC experiments / modeling goals and approaches at NETL
Approaches • Experimental Platform I. Plane-to-plane, high repetition rate nsec pulse discharge: large-volume, premixed, diffuse plasma chemical fuel oxidation and ignition at near-0-D conditions.
•Experimental Platform II. Point-to-point, single-pulse nsec pulse discharge: kinetics of energy transfer among excited species and radicals at high energy loadings per molecule
•Kinetic modeling. Integrated model of electric discharge dynamics, plasma kinetics / chemistry, and “conventional” hydrogen / hydrocarbon chemistry mechanism
Goals • Quantitative data in well-characterized plasma assisted combustion experiments: temperature, species number densities, vibrational state populations, electric field, electron density
• Quantify the effect of plasma generated species – radicals and excited states – on fuel oxidation, ignition, combustion, and flameholding
• Elucidate detailed kinetic mechanisms, develop predictive kinetic models of nonequilibrium plasma assisted combustion processes, assess and validate the models
If quasi-0-D geometry is assumed:
• Boltzmann equation for EEDF (two-term expansion, experimental cross sections): predict rates of electron impact excitation, dissociation, and ionization processes
• Charged species equations (ionization, recombination, attachment, detachment processes, ion-molecule reactions): predict electron density in plasma
• Excited neutral species equations (electron impact excitation, non-reactive and reactive quenching): predict contribution to radical species formation
• Master equation for N2 (X,v) populations; vibration-translation (V-T), vibration-vibration (V-V) processes, vibrational-chemistry (V-Chem) enhancement of reaction rates
• Neutral species reactions: fuel-air air chemistry, enhanced by radical production in plasma
Quasi-1-D and 2-D (plane or axisymmetric) geometry adds:
• Poisson equation for the electric field: predict electric field in plasma, cathode voltage fall
• Coupling between chemistry, transport processes (diffusion, conduction), and flow
• Nonequilibrium plasma model / code used as a starting point: non-PDPSIM (M. Kushner ), widely used in low-temperature plasma community, well-documented
Kinetic modeling of nonequilibrium fuel-air plasmas: brief model overview
Experimental Platform I: premixed, mildly preheated H2-air, CH4-air, C2H4-air, C3H8-air
• H2-air, C2H4-air, CH4-air, and C3H8-air at T0= 100-300° C, P=50-500 torr, ϕ=0.03-1.2 • Repetitive nanosecond pulse “bursts”: 20-25 kV peak, ~10-50 nsec, ν=10-40 kHz, 50-100 pulses • Ample optical access (LIF, Two-Photon LIF, psec CARS) for species and temperature measurements • OH LIF absolute calibration: adiabatic burner in Hencken burner, Rayleigh scattering
• Discharge dimensions 1 cm x 2 cm x 6 cm • Entire cell placed inside a tube furnace • Diffuse, volume filling plasma • Large-volume ignition (no propagating flame) • “Near 0-D” conditions
H2 – air, ϕ=0.3 T0=500 K, P=100 torr
C2H4 – air, ϕ=0.3 T0=500 K, P=100 torr
Pulse #10 Pulse #100
, pulse #10
[OH] on centerline after a 50-pulse burst, T0=500 K, P=100 torr: comparison with 0-D kinetic modeling (A. Konnov mechanism)
Good agreement for simple fuels, worse for more complex hydrocarbons
H2-air CH4-air
C2H4-air C3H8-air
2315 2320 2325 2330Raman Shift [cm-1]
DataTfit = 486K
Typical N2 psec CARS spectra and best fit Trot in air and H2-air
100-shot accumulation spectrum in 100 torr air, T0 = 500 K,
95% confidence interval ~15 K
2290 2300 2310 2320 2330 23400
5
10
15
20
25
30
Raman Shift [cm-1]
Sqrt[
Int]
[a.u
.]
ExperimentTfit = 1250 K
100-shot accumulation spectrum in 92 torr H2 –air during ignition, Tpeak= 1250 K
Also yields N2(v=1) level population, N2 vibrational temperature
N2(v=1)
N2(v=0)
[OH] (LIF) and psec CARS (T, Tv (N2)) measurements during plasma assisted ignition of H2-air
Threshold ignition temperature Tf ~ 700 K, lower than autoignition temperaure, Ta ~ 900 K
N2 vibrational temperature remains quite low
Use pin-to-pin discharge to enhance energy loading
This provides far more stringent test of the model
[OH] by OH LIF (Proc. Comb. Symp. 2013) 0-D model: good agreement with measured [OH] at the
end of the burst, during ignition
T0=500 K, P=92 torr, ϕ=0.4, v=10 kHz, 120 pulses
T, Tv (N2) by psec CARS (Comb. Flame 2013) Measured temperature in excellent agreement with 0-D
model predictions from previously published work
On-going high-pressure (~1 bar) experiments: use of liquid metal electrodes (Ga-In-St) in a “Wolverine” cell
Discharge test with salt water electrodes (cells on top and bottom of the channel)
P=40 torr, air, ν=10 kHz
Discharge channel, electrode cells filled with liquid metal (galinstan)
New high-pressure discharge channel, with electrode cells and preheating coil shown, gap L=5 mm
• Electrodes are fully encapsulated in quartz cells • No corona outside the channel, no damage of dielectric layers • No discharge pulse energy reduction and / or “drifting”
Nitrogen, P=100 torr
Experimental platform II: N2, air, H2-air, C2H4-air, initially at room temperature
10 mm
2 mm
Air, P=100 torr, 16 mJ/pulse
• No dielectric barrier: stable, diffuse, single filament discharge (~10 mm gap, ~2-3 mm diameter) at P=0.1-0.2 atm
• With dielectric barrier: volume-filling discharge (~1-2 mm gap, ~10 mm diameter) at P=0.5-1.0 atm
• High peak electric field (~10 kV/cm), electron density (~1014 - 1015 cm-3), electron temperature (~5 eV)
• High energy loading per molecule (up to ~ 0.2-0.3 eV/molecule), strong vibrational excitation, high concentrations of radical species (N2(X,v) molecules, N, O, H atoms, OH, NO …)
• Spatially resolved temperature, vibrational level populations, species concentrations measurements
• “Test bed” to study vibrational and electronic energy transfer, plasma chemical reactions in high energy loading, highly transient plasmas
Optical diagnostics data for insight into air and fuel-air plasma chemistry
Key plasma parameters controlling coupled energy and its partition among different channels (elastic, vibrational, electronic, dissociation, ionization):
• Electric field (CARS / 4-wave mixing)
• Electron density (Thomson scattering)
• Electron temperature (Thomson scattering)
Estimating E, ne, and Te from voltage and current results in significant uncertainty
Kinetic models need to predict them accurately, given applied voltage waveform
Key plasma parameters affecting plasma chemistry:
• Temperature and vibrational level populations (psec vibrational and rotational CARS, Rayleigh scattering, spontaneous Raman scattering)
• Radical species number densities (LIF, TALIF), including 1-D and 2-D imaging
Time-resolved and spatially-resolved data are highly desirable
Electric field measurements, plane-to-plane nsec pulse discharge in H2 (psec CARS / 4-wave mixing)
• H2, P=0.2-1.0 bar, discharge between two plane electrodes, 1 mm gap, pulse repetition rate 100 Hz • With or without dielectric barrier (glass plate 100 μm thick on one of the electrodes) • Without dielectric: single filament discharge; with dielectric: diffuse volume-filling discharge • Time resolution 0.2 nsec • Calibration: electrostatic field between two planes
H2, 430 torr Dielectric barrier: 100 μm glass plate
H2, 220 torr, no dielectric barrier
Experimental (RUB) and predicted (OSU) electric field in a plane-to-plane nsec pulse discharge in N2
• Experiment (Ruhr Universität Bochum, Germany, nsec pulse 4-wave mixing): N2, P=0.25 bar, 1.2 mm gap, pulse repetition rate 2 kHz
• Field in plasma before breakdown follows applied voltage, E=U/d • Field in plasma after breakdown is significantly lower, significant cathode voltage fall • Model predictions for time-resolved E-field, pulse current: good agreement with experiment
0.25 bar
Preliminary results: psec CARS E-field in a nsec pulse discharge between two plane metal electrodes
• H2, discharge between two plane electrodes, 1 mm gap, pulse repetition rate 10 Hz
• With or without dielectric barrier (glass plate 100 μm thick on one of the electrodes)
• Without dielectric: single filament discharge • With dielectric: diffuse volume-filling discharge • Time resolution 0.2 nsec (0.2 nsec “time bins” are used) • Absolute calibration: electrostatic field between two planes • Kinetic modeling calculations are underway
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1-45
-40
-35
-30
-25
-20
-15
-10
-5
0
5
Time (µsec)
Elec
tric
Fiel
d (k
V/cm
)
Experimental ResultsHV Probe
H2, 430 torr diffuse
-1 -0.5 0 0.5 1-12
-10
-8
-6
-4
-2
0
2
Time (µsec)
Elec
tric
Fiel
d (k
V/cm
)
Negative Polarity Calibration
Experimental ResultsHV Probe
-600 -400 -200 0 200-5000
0
5000
10000
15000
20000
Time (nanoseconds)
V/cm
p
Results - 0.2 nsec binsAverage HV Probe Results
H2, 220 torr filament
Helium, 200 torr, 10 mm gap, camera gate 150 nsec Left: single pulse; right: 100-pulse average
Thomson scattering electron density measurements in diffuse filament discharges in He, H2-He, O2-He
Main pulse
“Afterpulse”
Absolute calibration of Thomson scattering data using N2 rotational Raman spectra
No discharge Used for Thomson scattering calibration Laser energy 500 mJ/pulse 6 minutes accumulation time
05000
1000015000200002500030000
526 528 530 532 534 536 538
Inte
nsity
(a.u
.)
Wavelength(nm)
02000400060008000
1000012000
525 530 535 540Wavelength (nm)
t=10 ns
GaussianFit
0
20000
40000
60000
80000
525 530 535 540Wavelength (nm)
t=90 ns
Gaussian Fit
Sample Thomson scattering spectra during and after discharge pulse (t=0-250 nsec)
• Helium, 200 torr
• Peak voltage 7 kV
• Peak current 60 A
• Coupled energy ~18 mJ/pulse
• Pulse repetition rate 90 Hz
• t=0 beginning of pulse current rise
• Laser energy 500 mJ/pulse
• 6 minutes accumulation (~3200 shots)
Rayleigh scattering blocked
Experimental and predicted (2-D model) electron density and electron temperature in helium
• Breakdown may be not fully resolved in the experiment (time resolution ±15 nsec) • High electron temperature (up to ~5 eV) during breakdown onset • Peak ne during the pulse, ne ~ 3·1015 cm-3, followed by rapid decay in the afterglow • “Residual” Te in the afterglow (maintained by superelastic processes) Te ~ 0.3-0.4 eV) • Ongoing work: electron density measurements in molecular gases (H2, O2, N2, air) • Rotational Raman spectra are “in the way” (except for H2), but • High electron density, ~1014 ~ 1015 cm-3 helps inferring ne , Te from underlying “envelope”
Combined Thomson / Raman spectra in nsec pulse discharges in O2-He mixtures
• Raman spectrum is taken after electron-ion recombination has occurred (5 μs after the pulse) and subtracted
• Temperature can be inferred from the Raman spectra
10% O2 in He, P=100 torr ne = 6·1013 cm-3, Te =1.7 eV
Temperature, N2(X,v) populations measurements: psec BoxCARS with broadband dye mixture
Very broadband pyrromethene dye mixture:
access to high vibrational levels
-0.4 -0.2 0 0.2 0.40
0.5
1
1.5
2
2.5
3
Distance [mm]
Inte
grat
ed S
igna
l Int
ensi
ty [a
u]
95% regionDataGaussianFit
Interrogation volume
CARS signal beam Folded Box-CARS
High spatial resolution: 95% of signal generated over ~0.5 mm
Typical Psec CARS Spectra, 100 Torr N2 (Normalized to v=0, corrected for dye laser spectral profile)
v=0 ↓
v=3 ↓
v=6 ↓ v=9
↓
100 laser “shot” averaged spectra vs. time after rising edge of current pulse
Vibrational level populations inference: least squares fitting to Voigt line shape
10 mm
2 mm
N2, P=100 torr, 5 mJ/pulse, 100 μs delay
Radial distribution of “first level” vibrational temperature
•Strong vibrational disequilibrium, up to Tv01(N2) = 3000 K ( ≈ 0.8 vibrational quanta per molecule), moderate translational/rotational mode temperatureTrot = 300-800 K
•N2(X) vibrational levels up to v=9 are detected
•First level vibrational temperature keeps rising after the discharge pulse: V-V energy transfer from higher vibrational levels, e.g. N2(v=2) + N2(v=0) → N2(v=1) + N2(v=1)
•Gradual relaxation on long time scale: V-T relaxation, diffusion out of the filament region
Results: psec CARS measurements in air, during and after nsec discharge pulse
Air, P=100 torr, 16 mJ/pulse
[ ]1001 ln nn
T vv
θ=
“Rapid” and “Slow” Heating in Air and Nitrogen
• “Rapid” heating in N2 and air: N2(A) + N2(A) → N2(B,v) + N2(X,v) (nitrogen) N2(A,B,C,a) + O2 → N2(X,v) + O + O (air)
• “Slow” heating in air (nearly absent in N2), V-T relaxation by O atoms: N2(X,v) + O → N2(X,v-1) + O
Compression waves generated by the filament at 1-10 μsec (frames are 1 μsec apart)
•Psec CARS: T, Tv(N2), [N2(X,v)]
• TALIF, LIF: absolute N, O, NO number densities
• Vibrational excitation and temperature rise predicted accurately
• N2(X 1Σg+, v) + O → NO + N channel appears unlikely: N2(X)
vibrational excitation (measured by CARS) is fairly weak
• [NO] reproduced only when formation processes via multiple N2 excited electronic states, N2
* + O → NO + N, are incorporated
• [NO] reduction is nearly the same as initial [N], ~1015 cm-3: NO+ N → N2(v) + O reaction
Air, P=100 torr: “Full set” of data (T, [N2(v)], [N], [O], [NO], NO PLIF)
NO PLIF image 10 µs after discharge pulse
Air, P=100 torr, 4 mJ/pulse
N, NO in air, H2-air, P=40 Torr N atoms “scavenge” OH radicals
Adding H2 increases NO decay time by a factor of ~100, reduces N atoms Similar effect in C2H4 - air
Air ϕ = 0.15 ϕ = 0.42
Longer NO decay primarily due to reaction N + OH → NO + H
HV
GND
HV
GND
Ar:O2:H2=80:20:2
Laser beam
Flow
P=40 torr ν=100 kHz
50 pulses
~100 ns long each 10 µs apart
Electrode gap: 11.7 mm, laser beam 4.7 mm from high voltage electrode
500-pulse accumulation
Ar:H2=100:2
Spatially resolved measurements of H2-O2-Ar plasma chemistry, point-to-point discharge
2 mm
500 µs
1 ms
200 µs
100 µs
55 µs
15 µs
Flow H TALIF signal OH LIF signal
H atom TALIF and OH LIF after 50-pulse burst, 2% H2 - 20% O2 – Ar, P=40 torr
5 mm 5 mm
HV
GND
5th pulse
OH PLIF image 2% H2 - 20% O2 - Ar
2 mm
50th pulse
OH PLIF image 2% H2 - 20% O2 - Ar
HV
GND 2 mm
100 µs after
OH PLIF image 2% H2 - 20% O2 - Ar
HV
GND 2 mm
500 µs after
OH PLIF image 2% H2 - 20% O2 - Ar
HV
GND 2 mm
T (Rayleigh scattering), absolute [H] (TALIF), and [OH] (PLIF ) after 50-pulse burst, 2% H2 - 20% O2 - Ar, P=40 torr
Hot central region: chain branching reactions dominate OH production
H + O2 → OH + O ; O + H2 → OH + H
Lower temperature peripheral region: predominant OH accumulation
H radial diffusion ; H + O2 + M → HO2 ; H + HO2 → OH + OH
• Growing body of time-resolved, spatially-resolved data characterizing pulsed, high-pressure fuel-air plasmas
• Measurements and predictions of electric field, electron density, temperature, and N2(X,v) populations are necessary for insight into discharge energy partition
• Measurements and predictions of excited electronic states of N2* and key radicals (O, H, OH,
and NO) are critical for quantifying their effect on fuel-air plasma chemistry
• “Minimum” data set for validation of low-temperature plasma assisted combustion chemistry mechanism: time-resolved temperature, N2 vibrational temperature, and key radical concentrations during a repetitively pulsed plasma-enhanced ignition process
• Parameters used for conventional combustion mechanism validation (such as ignition delay time, laminar flame speed) are insufficient by far
• Kinetic sensitivity analysis: identify reduced reaction mechanism for coupled discharge dynamics / molecular energy transfer / plasma chemistry kinetic model, incorporating a wide range of time scales and realistic geometry
Summary / Future Work
Unresolved Issues
• Effect of “rapid” heating compared to reactions of plasma-generated radicals: at what conditions (pressure and temperature) does “rapid” heating become dominant effect in transient fuel-air plasmas, compared to low-temperature radical species chemistry?
• Effect of reactions of vibrationally excited molecules, compared to reactions of plasma-generated radicals? Do reactions such as N2(X1Σ,v) + O → NO + N and N2(X1Σ,v=1) + HO2 → N2(X1Σ,v=0) + HO2(ν2+ν3) → N2 + H + O2 really matter and if so, at what conditions?
• Effect of fuel molecular structure on plasma-generated radicals chemistry. Is there a difference between plasma assisted combustion of low octane number fuels (exhibiting low-temperature cool flame chemistry) vs. high octane number fuels (cool flames are not observed)?
• Dynamic effect of plasma on non-premixed turbulent flames: preventing local extinction by producing radicals where F/A ratio, temperature, or pressure becomes too low to sustain combustion. Need high frame rate (~10 kHz) imaging of temperature and radical species concentrations fields.
• Plasma assisted combustion in non-premixed compressible flows: understand coupling between electric discharge dynamics, fuel-air mixing, and combustion instability development.
AFOSR MURI “Fundamental Mechanisms, Predictive Modeling, and Novel Aerospace Applications of Plasma Assisted Combustion” US DOE Plasma Science Center “Predictive Control of Plasma Kinetics: Multi-Phase and Bounded Systems” NSF "Kinetics of Non-Equilibrium Fast Ionization Wave Plasmas in Gas Phase and Gas-Liquid Interface“
Acknowledgments